|Hi Uday and Georg,|
long ago we were discussing about implementing pressure for MMM2D and/or ELC. First of all, the simple dU/dV approach only works for 3d systems and only for the isotropic pressure, so one actually needs to compute the full pressure tensor.
In principle one can derive that from the MMM2D formulas, but practically that is quite tedious: at the heart of MMM2D/ELC there are trigonometric addition theorems. When deriving the cross terms of the pressure tensor, you get so many contributions that we found it impossible to implement the algorithm by hand, and writing a program to generate the code seemed somewhat overdoing it. Things got even more difficult because the current MMM2D/ELC also handles dielectric contrasts, which made the formulas even more awkward...
this just means that when you run 'analyze pressure total', the resulting value does not contain the pressure contribution due to the electrostatic interactions. There is nothing wrong with the dynamics of the system, this only influences this calculated observable.
Do you need to know the pressure? I assume it's not a problem in principle to include this contribution in the calculation, but so far no one has implemented the necessary code.